PASFRG Minutes   2/27/98

We are  STILL urgently in need of volunteers to take on one of the categories of
requirements and quantify and expand on the initial rather terse statements.

Paul Lebrun gave a presentation about the Nirvana suite of tools; his slides are
linked from the PASFRG home page.  A couple of important points were raised in
discussion that apply to other products as well:
1) Moving code back and forth between an off-line batch environment and the
interactive (interpreted) analysis environment.  We need to understand this
requirement precisely, as it might rule out otherwise attractive solutions like
Python (or other) languages for scripting.
2) Scaling limitations: We need to understand where each product breaks down as
data sample sizes get larger and larger.

John Marraffino led a discussion of PAW limitations.  Many opinions were
expressed.  Concrete iussues raised included:
1) Problems with COMIS (especially double precision)
2) Lack of a debugger with COMIS
3) Lack of support (esp for porting to new operating
systems/chips/architectures)
4) 64 bit operating systems will require a major rewrite of ZEBRA
5) New data types a problem
6) HIGZ limitations (can't use modern fonts)
7) Many old nagging problems with private work arounds (only one explicitly
mentioned was file size limitations)
8) Performance limitations

It is unclear whether or not any of these by themselves are fatal, or whether
the accumulated set of problems is fatal.  John will try to collect and quantify
these concerns, members of the working group should send him comments

Watch the agendas on the web page for which products will be discussed when.

Future Meetings (all in FCC2A):

Fri Mar 13, 8:30AM